基于Kohonen网络的软件可靠性模型选择

Selection of software reliability model based on Kohonen network

软件可靠性模型是软件可靠性工程的一个重要方面。现在还没有一个通用的模型,模型选择问题已成为模型研究的重点。运用聚类思想对软件可靠性模型的选择进行研究。在对软件失效数据进行编码的基础上,采用Kohonen神经网络对其进行聚类分析,从而实现了可靠性模型的选择。最后通过仿真测试,证明了此方法的有效性和可行性。

Software reliability model selection is a important part of software reliability engineering. Now there is no model which is suitable for all users. Moreover, the accuracy of reliability measure for using different model is different. Model selection has been an emphasis in the research of model. In this paper, the clustering analysis was used as a tool for software reliability selection. On the basis of Failure data coded, Kohonen nets were used as a tool for clustering analysis. Twenty data sets were used to demonstrate the approach and efficiency.